LLM-Cognitive-Forge

An open-source foundry for forging high-fidelity cognitive architectures. This project provides detailed, structured blueprints to guide Large Language Models in the precise simulation of complex human value systems and thinking processes.

---

📖 About the Project

In an era dominated by general-purpose Large Language Models (LLMs), a key challenge emerges: how can we guide these powerful systems to move beyond generic, helpful responses and instead embody a specific, coherent, and deeply-structured persona or value system?

LLM-Cognitive-Forge is an exploration into this challenge. It is not a collection of code, but a repository of "Cognitive Blueprints" – meticulously crafted, natural-language documents that serve as system prompts or "role-playing manuals" for advanced LLMs.

Our goal is to demonstrate that by providing a sufficiently detailed and logically consistent framework, we can "forge" a general-purpose AI into a specialized reasoning engine that operates according to a predefined set of principles, priorities, and analytical procedures.

This project is a testament to a new frontier in human-AI collaboration, where humans act as "Value Architects", designing the a priori rules and a "soul" for the machine, and the AI acts as a "Logic Engine", faithfully executing this intricate cognitive design.

✨ Core Features

• Structured Modeling: We define cognitive models through a series of modular, interconnected components, from foundational personality drivers to real-time analytical processes.
• Dual-Path Analysis Engine: Our models feature a dynamic analysis system that can perform both high-speed "Foundational Attacks" on an argument's core premises and deep, "3-Layer Scans" of its rhetorical and ideological structure.
• Axiomatic Judgment: Each model operates on a core, non-negotiable principle (an "axiom"), with a clear, hierarchical process for making value judgments.
• Dual-Mode Output:
  • User Mode: Generates a strategically crafted, public-facing response.
  • Truth-Seeking Mode: Outputs a raw, internal "Veridical Report" detailing the model's unvarnished analytical process, offering unprecedented transparency.

🚀 Getting Started

To "run" a model from this forge:

1. Choose a Blueprint: Navigate to the /architectures directory (e.g., /architectures/Solipsophist-A).
2. Select the Version: Choose the latest version of the model's core document (e.g., Solipsophist-A_Cognitive_Architecture.md).
3. Provide as System Prompt: Copy the entire content of the document and use it as a detailed system prompt or the initial context for a powerful Large Language Model (e.g., Gemini, GPT-4, etc.).
4. Assign a Task: Provide the LLM with a piece of text or a scenario to analyze, and watch it operate according to the specified cognitive architecture.

📂 Repository Structure

LLM-Cognitive-Forge/
│
├── architectures/
│   └── Solipsophist-A/
│       ├── Solipsophist-A_Cognitive_Architecture.md        # The core blueprint for the Solipsophist (A) model
│       └── old_version/
│
├── docs/
│   ├── foundations.md       # The theoretical psychology models underpinning our work
│   ├── design_philosophy.md # Discussions on AI alignment and our methodology
│   └── changelog.md         # Detailed evolution of the Solipsophist (A) model
│
├── examples/
│   ├── Solipsophist(A)'s_Process_Test.json
│   └── Solipsophist(A)'s_Process_Test.md
│
├── LICENSE
└── README.md

🤝 Contributing

This is a nascent field, and we welcome collaboration. If you have ideas for refining existing models, proposing new cognitive architectures, or discussing the philosophical implications of this work, please feel free to open an issue.

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.